There's an entire course 10-725 on this topic of convex optimization; it seems that if given a pretty simple graph, there's a lot of good algorithms and theorems about finding the minimum.
keenan
@Azure Yep, if you can cast your problem as a convex optimization problem, you're usually in pretty good shape. However, there are still convex problems that are expensive to solve for large problem sizes (e.g., semidefinite programs), and there are even convex problems that are not computationally easy to solve (though quite rare in graphics).
There's an entire course 10-725 on this topic of convex optimization; it seems that if given a pretty simple graph, there's a lot of good algorithms and theorems about finding the minimum.
@Azure Yep, if you can cast your problem as a convex optimization problem, you're usually in pretty good shape. However, there are still convex problems that are expensive to solve for large problem sizes (e.g., semidefinite programs), and there are even convex problems that are not computationally easy to solve (though quite rare in graphics).